JP5531620B2 - Semiconductor device - Google Patents

Description

  The present invention relates to a structure of a semiconductor device, particularly a Schottky barrier diode (hereinafter abbreviated as SBD) having a trench structure.

    FIG. 15 shows a cross-sectional structure of a Trench MOS Barrier Schottky diode (hereinafter referred to as a TMBS diode) which is an SBD having a trench structure. An n-type drift layer 2 and an anode electrode 3 that forms a Schottky junction with the n-type drift layer 2 are formed on the surface of the TMBS diode. In addition, active portion trenches 12 whose inner walls are covered with the oxide film 11 are formed at equal intervals in the active portion 21 that becomes a current path during conduction. The inside of the active portion trench 12 is filled with the same metal as the anode electrode or a conductor such as conductive polysilicon. When a reverse bias voltage is applied to the TMBS diode, a depletion layer spreads from the Schottky junction formed by the anode electrode 3 and the n-type drift layer 2. When the depth of the depletion layer becomes deeper than the bottom of the trench, equipotential lines are concentrated on the oxide film 11 formed at the bottom of the trench and having a dielectric constant lower than that of silicon. Strength decreases. As a result, not only the leakage current is reduced by suppressing the phenomenon of barrier height reduction, but also the applied voltage can be shared with the oxide film. Therefore, the withstand voltage can be improved over the conventional SBD due to the well-known RESURF (Reduced Surface Electric Field) effect. Further, since the doping concentration of the n-type drift layer 2 can be increased by the RESURF effect, a lower ON resistance can be realized with a leakage current equivalent to that of the related art, despite a high breakdown voltage.

  In the peripheral region of the anode electrode 3, a breakdown voltage structure portion 22 (an electric field relaxation region formed on the outer peripheral side of the active portion 21) is formed. A trench is formed at an end portion of the anode electrode 3 (hereinafter referred to as an active end portion 19), and the anode electrode 3 is terminated at an upper portion of the polysilicon 13 embedded in the trench. Hereinafter, the trench formed in the active end portion 19 is referred to as an end portion trench 7. A p-type guard ring layer 5 is formed between the active portion trench 12 and the end trench 7 and is connected to the anode electrode 3. Although not shown, there is a structure in which the p-type guard ring layer 5 is further omitted. Here, when the p-type guard ring layer 5 is not particularly provided, the active end 19 is terminated by the upper surface of the polysilicon 13 embedded in the end trench 7 so that the active end 19 is an n-type drift layer. Do not touch 2 directly. By doing so, in the n-type drift layer 2 in the vicinity of the active end portion 19, the electric field strength is prevented from locally increasing when a reverse bias is applied.

  Next, the processing method of the edge part of the longitudinal direction of the active part trench 12 and the positional relationship with the edge part trench 7 are demonstrated. FIGS. 16A and 16B are plan views showing the positional relationship between the active trench 12 and the end trench 7 according to the conventional concept. Here, description of the side wall oxide films formed in these trenches and the polysilicon 13 embedded in the trenches is omitted, and these are included in FIGS. 16-1 and 16-2. In FIG. 16A, the active portion trench 12 is processed so that another active portion trench 12 adjacent to the end portion in the longitudinal direction of the active portion trench 12 is connected. In other words, the active portion trenches 12 have a donut shape and are adjacent to each other. Such end treatment is a method often found in the layout of a plurality of elongated trench gates in a trench gate MOSFET or IGBT. Further, as shown in FIG. 16B, there is also a method of terminating the active portion trench 12 as it is at the end portion in the longitudinal direction. In this case, the end portion of the active portion trench 12 is terminated in a semicircular shape whose diameter is the width of the trench.

JP 2002-50773 A

  However, it has been found that such a conventional structure has the following problems. When a high reverse bias is applied between the anode electrode 3 and the cathode electrode 4, the electric field is concentrated near the bottom on the outer peripheral side of the end trench 7 of the active portion 21. Therefore, the avalanche breakdown occurs not at the active portion 21 but at the position Q described in FIG. Since the avalanche current mainly flows through the breakdown voltage structure 22 rather than the active portion 21, the breakdown voltage decreases as a result of the avalanche current concentrated on the breakdown voltage structure 22.

  As a solution, there is a method of thickening the oxide film 11 on the trench side wall. However, for example, in order to form the oxide film 11 having a thickness of 5000 mm or more, the gas flow must be controlled while maintaining a temperature of 1000 ° C. or higher for a long time in the oxidation process, and the processing process itself becomes difficult. . Further, equipotential lines concentrate on the oxide film 11 having a low dielectric constant when a high reverse bias voltage is applied. As a result, the above-described good resurfing effect cannot be obtained, and the effect of increasing the breakdown voltage and reducing the leakage current, which are the merits of the TMBS diode, are reduced.

    Further, when the p-type guard ring layer 5 is in contact with the anode electrode 3, there are the following problems. When the value of the forward bias voltage during the on operation exceeds the built-in potential of the pn junction formed by the p-type guard ring layer 5 and the n-type drift layer 2, a forward bias is applied to the pn junction, and minority carriers ( Holes) are injected into the n-type drift layer 2. For this reason, when switching to the off state, the accumulated minority carriers are swept out, so that the reverse recovery time becomes extremely long. Therefore, the p-type guard ring in contact with the n-type drift layer 2 becomes a factor that hinders high-speed operation, which is one of the advantages of the TMBS diode.

  Also, there is a problem with the conventional method for treating the end of the active portion trench 12 in the longitudinal direction. When the termination process shown in FIG. 16A is applied to a TMBS diode, the electric field strength is concentrated at a position M shown in FIG. That is, since the end of the active portion trench 12 is curved with a certain radius of curvature, the equipotential lines are also reflected toward the outside of the donut shape reflecting this shape, and thus curved. Then, since the electric field strength is proportional to the spatial gradient of the electrostatic potential, the electric field strength increases from the inside of the active portion trench 12 having a linear stripe shape. Further, the electric field strength is maximum at the position M farthest from the adjacent trenches including the end trench 7. Therefore, an avalanche is likely to occur, and a leakage current also increases at the position M due to a well-known phenomenon of lowering the Schottky barrier. Further, in the method shown in FIG. 16B, the end of the active portion trench 12 is terminated in a semicircular shape whose diameter is the width of the trench, but its radius of curvature is extremely small. Then, as described above, the electric field strength is greatly increased at the end of the trench. Further, the semiconductor layer (n-type drift layer 2) around the active portion trench 12, the oxide film 11 formed on the side wall of the active portion trench 12 by thermal oxidation or the like, and the polysilicon 13 embedded inside the oxide film 11 During this time, stress is generated (see FIG. 15). Since this stress increases with a decrease in the radius of curvature of the trench edge, the crack 14 frequently occurs in the mesa region 18 as shown in FIG.

  The present invention has been made in view of such conventional problems, and in the trench formed at the end of the active portion, it is easy to manufacture while relaxing the electric field strength concentrated on the bottom of the outer periphery of the trench. An object of the present invention is to provide a semiconductor device having a low leakage current, a high breakdown voltage, and a small amount of minority carrier injection.

A cathode layer made of a first conductivity type semiconductor substrate, and a drift layer made of a first conductivity type semiconductor substrate having a lower concentration than the cathode layer is provided on one main surface of the cathode layer, and on the upper surface of the drift layer At least one first trench and an end trench surrounding the first trench are provided, and the first conductor and the end trench are filled with a first conductor via an insulating film, A semiconductor in which an anode electrode is provided on the upper surface of the drift layer so as to be in contact with the conductor and form a Schottky junction with the drift layer, and a cathode electrode is provided on the other main surface of the cathode layer In the apparatus, the outer peripheral end of the anode electrode is in contact with the first conductor of the end trench, and a field plate is provided apart from the anode electrode. A second trench is provided so as to surround the end trench apart from the end trench, and a second conductor is embedded in the second trench via an insulating film, and the field trench plate is a semiconductor device characterized in that it connects to the surface the same potential of the drift layer in the mesa regions between the second conductor and said end portion and second trenches.

The structure of the semiconductor device in the above invention has the following characteristics with respect to the active portion and the breakdown voltage structure portion of the TMBS diode.
(1) With respect to an end trench provided so as to surround one or more first trenches formed in the active portion, an end portion of the anode electrode is in contact with a conductor formed inside the end trench. Yes.

(2) A second trench is formed on the outer peripheral side of the end trench so as to be separated from and surround the end trench.
(3) On the outer periphery of the anode electrode, a field plate separated from the anode electrode has a part of the surface of the mesa region of the n-type drift layer between the end trench and the second trench, It is formed in contact with both the conductor formed inside the second trench.

  With the above configuration, when a reverse bias voltage is applied to the junction between the anode electrode and the n-type drift layer, the potential in the vicinity of the second trench becomes higher than that of the anode electrode. At this time, in the mesa region between the end trench and the second trench, the potential at the location where the surface of the n-type drift layer in the mesa region is in contact with the field plate is the conductor inside the second trench. And the same potential. As a result, the depletion layer extending from the Schottky junction between the anode electrode and the n-type drift layer is pulled to the potential of the field plate. Thus, near the surface of the pressure-resistant structure portion, the depletion layer is likely to spread in a direction parallel to the surface, and the electric field strength near the bottom of the end trench can be relaxed.

More preferable means for the above invention will be described.
A distance W1 from the outer peripheral side wall of the end trench to a chip inner peripheral end (hereinafter referred to as a position P) of a region where the field plate and the drift layer are in contact with each other is It is preferable that the distance is smaller than the distance W2 to the inner peripheral end of the second trench.

  In this case, since the distance from the outer peripheral side end portion of the trench at the active end portion to the position P is narrow, the depletion layer spreading at the time of reverse bias is pulled more strongly by the potential of the field plate. For this reason, since the depletion layer can spread toward the second trench with a low reverse bias voltage, the effect of relaxing the electric field strength in the vicinity of the bottom of the trench at the active end is enhanced. As a result, it is possible to prevent the avalanche current from being concentrated on the breakdown voltage structure.

The width of the end trench is preferably larger than the width of the first trench.
It is desirable to make the width of the first trench, which becomes an invalid region when conducting, as small as possible. On the other hand, in the end trench, the end portion of the anode electrode must be terminated in a region of a conductor such as polysilicon embedded in the end trench. Therefore, if the width of the end trench is made larger than the width of the first trench, the anode electrode can be stably terminated.

  Further, a radius that is disposed between the first trench and the end trench, the length of the straight portion is shorter than the length of the first trench, and both ends are smaller than the radius of curvature of the end trench 7. It is preferable to have a third trench connected to the first trench provided at the end of the first trench.

  In this case, normally, the end portion in the longitudinal direction of the first trench sidewall formed in the active portion is curved with respect to the upper surface of the chip and has a small radius of curvature, so that the depletion layer expands when a reverse bias is applied. Sometimes the electric field strength increases. Therefore, by adopting the above configuration, there is no end portion in the longitudinal direction of the first trench, so that the concentration of the electric field strength as described above does not occur.

  Alternatively, the plurality of first trenches formed in the active portion have a donut shape on the upper surface of the drift layer, and a geometric center thereof is the innermost circumference of the anode electrode in the first trench. It is preferable to be located inside a donut-shaped trench having a minimum diameter formed in.

In this case, the first trench formed in the active part has a donut shape. For this reason, the first trench has no end in the longitudinal direction of the side wall thereof. Further, the geometric gravity center position of the donut-shaped first trench on the chip surface is provided so as to be inside the innermost first trench formed near the center of the active portion. Therefore, there is no concentration of electric field strength in the vicinity of the end portion in the longitudinal direction as described above, and it is possible to prevent a decrease in breakdown voltage due to the concentration of the electric field strength.
Further, a second conductivity type floating layer connected to both or one of the end trench and the second trench and to the field plate and formed on the upper surface of the drift layer includes the anode electrode. And the depth of the floating layer from the upper surface of the drift layer may be deeper than the depth of either or both of the end trench and the second trench.

  In this case, since the depletion layer that spreads at the time of reverse bias can reach the floating layer before the second trench, the effect of pulling the depletion layer further to the outer peripheral side becomes stronger. As a result, the electric field strength in the vicinity of the bottom of the end trench and the second trench is further relaxed. In addition, since the p-type floating layer is not in contact with the anode electrode, the electric field strength can be relaxed without injecting holes, which are minority carriers, into the drift layer. Furthermore, since the depletion layer extends from the pn junction of the floating layer because the junction depth of the floating layer becomes deeper than the first or second trench, the bottom of the trench located at a position shallower than the junction depth. There is almost no depletion layer. For this reason, the electric field strength at the bottom of the trench hardly increases, and the breakdown voltage can be determined only by the structure of the active region.

The floating layer may be in contact with the second trench.
In this way, the depletion layer is more likely to spread around the chip periphery, and as a result, the breakdown voltage value including the breakdown voltage structure portion can be made larger than the breakdown voltage value of only the active portion.

  In addition, the surface of the drift layer sandwiched between the end trench and the second trench has a concentration higher than the concentration of the drift layer and either or both of the end trench and the second trench. It is preferable that the first conductivity type surface layer shallower than either one is formed, and the maximum concentration of the surface layer is not less than the value indicated by the drift layer and 10 times the value indicated by the drift layer. It may be the following.

  When external charges enter the breakdown voltage structure, the surface layer makes it difficult for the surface of the mesa region 18 to be charged, and hole channel formation or electric field strength distribution does not occur and breakdown voltage or leakage current is stable. To do.

  As described above, according to the present invention, manufacturing is easy while relaxing the electric field strength concentrated on the bottom of the outer periphery of the trench formed at the end of the active portion, the breakdown voltage is high with low leakage current, and minority carriers. It is possible to realize a semiconductor device in which the injection of silicon is suppressed.

It is principal part sectional drawing of the semiconductor device concerning embodiment of this invention. It is principal part sectional drawing of the semiconductor device concerning embodiment of this invention. It is principal part sectional drawing of the semiconductor device concerning embodiment of this invention. It is principal part sectional drawing of the semiconductor device concerning embodiment of this invention. It is principal part characteristic sectional drawing of the semiconductor device of a prior art example. It is principal part characteristic sectional drawing of the semiconductor device concerning embodiment of this invention. It is principal part characteristic sectional drawing of the semiconductor device concerning embodiment of this invention. It is a characteristic view which showed the electrical characteristic in the principal part depth direction of the semiconductor device concerning embodiment of this invention. It is a characteristic view which showed the electrical characteristic in the principal part depth direction of the semiconductor device concerning embodiment of this invention. It is principal part sectional drawing of the semiconductor device concerning embodiment of a prior art example. FIG. 10 is a characteristic relationship diagram showing electrical characteristics of a semiconductor device according to an embodiment of the present invention and a conventional example. It is a characteristic relation figure of the semiconductor device concerning an embodiment of this invention. It is the characteristic view which showed the current-voltage characteristic of the semiconductor device concerning embodiment of this invention and a prior art example. FIG. 5 is a characteristic relationship diagram showing electrical characteristics of the semiconductor device according to the embodiment of the present invention. It is a principal part top view of the semiconductor device concerning embodiment of this invention. It is a principal part perspective view of the semiconductor device concerning embodiment of this invention. It is a principal part top view of the semiconductor device concerning embodiment of this invention. It is a principal part perspective view of the semiconductor device concerning embodiment of this invention. It is principal part sectional drawing of the semiconductor device concerning embodiment of this invention. It is principal part sectional drawing of the semiconductor device concerning embodiment of this invention. It is principal part sectional drawing of the semiconductor device concerning embodiment of this invention. It is principal part sectional drawing of the semiconductor device of a prior art example. It is a principal part top view of the semiconductor device concerning a prior art example. It is a principal part top view of the semiconductor device concerning a prior art example.

  Hereinafter, the first conductivity type is n-type and the second conductivity type is p-type, but the present invention operates similarly even if the n-type and p-type are interchanged. In the following embodiments, the TMBS diode, which is a semiconductor device, also uses the expression element or chip, but shows the same object.

A TMBS diode according to a first embodiment of the present invention will be described with reference to FIG.
An n-type drift layer 2 having a lower concentration than the n-type semiconductor substrate 1 is formed on the upper surface of the n-type semiconductor substrate 1 in order to keep the breakdown voltage of the element high. On the upper surface of the n-type drift layer 2, an active portion 21 and a breakdown voltage structure portion 22 which are main paths through which a current flows are formed. The breakdown voltage structure portion 22 is a region for relaxing the electric field strength concentrated on the outer peripheral side of the active portion 21 when a reverse bias is applied to the element and the depletion layer spreads toward the outer peripheral portion of the chip. On the upper surface of the chip in the active portion 21, trenches are arranged with a constant period (hereinafter referred to as the active portion trench 12). An oxide film 11 is formed on the side wall of the active trench 12, and conductive polysilicon 13 is embedded in the oxide film 11. The oxide film 11 insulates the n-type drift layer 2 from the polysilicon 13. An anode electrode 3 is formed on the upper surface of the active portion 21 so as to form a Schottky junction with the n-type drift layer 2. At this time, the width of the mesa portion between the adjacent active portion trenches 12 is that of the built-in depletion layer extending from the side wall of the Schottky junction 16 and the adjacent active portion trenches 12 on both sides in the thermal equilibrium state to the n-type drift layer 2. It is preferably narrower than twice the width Wbi. In this way, when a reverse bias voltage is applied, the depletion layer extending from the sidewalls of the Schottky junction 16 and the adjacent active trench 12 on both sides is immediately pinched off (the depletion layers extending from different directions are combined to form one Can spread like a depletion layer). As a result, since the electric field strength of the Schottky junction 16 can be suppressed to be small, the phenomenon of reducing the Schottky barrier height hardly occurs, and an increase in leakage current can be suppressed. The anode electrode 3 is also in contact with the polysilicon 13 and is in ohmic contact with the polysilicon 13. If the active end portion 19 is defined more precisely, the active end portion 19 is an end portion of a region where the anode electrode 3 is in contact with the n-type drift layer 2 or the polysilicon 13. In the first embodiment of the present invention, the active end 19 is always terminated inside the polysilicon 13 so that the active end 19 is not in direct contact with the n-type drift layer 2. By doing so, in the n-type drift layer 2 in the vicinity of the active end portion 19, the electric field strength is prevented from locally increasing when a reverse bias is applied. A trench that is formed in the active end 19 and terminates the anode electrode is referred to as an end trench 7. An interlayer insulating film 6 is formed above the end trench 7 and extends in the chip outer peripheral direction. The anode electrode 3 extends in the chip outer peripheral direction on the upper surface of the interlayer insulating film 6 and terminates on the upper surface of the interlayer insulating film 6. On the other hand, a guard trench 8 is formed in the pressure-resistant structure 22 on the chip outer peripheral side with respect to the end trench 7 so as to be separated from the end trench 7. Inside the guard trench 8, like the active portion trench 12, an oxide film is formed on the trench side wall, and conductive polysilicon 13 is embedded. Further, a conductive field plate 9 is formed so as to contact the polysilicon 13 embedded in the guard trench 8. The field plate 9 is formed not only in the polysilicon 13 inside the guard trench 8 but also in the mesa region 18 of the n-type drift layer 2 between the end trench 7 and the guard trench 8 and the opening on the upper surface of the mesa region 18. Connected.

  With the above configuration, when a reverse bias voltage is applied to the junction between the anode electrode 3 and the n-type drift layer 2, the potential in the vicinity of the guard trench 8 becomes higher than that of the anode electrode 3. At this time, in the mesa region 18 between the end trench 7 and the guard trench 8, the potential where the n-type drift layer 2 is in contact with the field plate 9 is the same as that of the conductive polysilicon 13 inside the guard trench 8. Become. As a result, the depletion layer extending from the Schottky junction 16 between the anode electrode 3 and the n-type drift layer 2 is pulled to the potential of the field plate 9. Thus, also in the breakdown voltage structure 22, the depletion layer is likely to spread toward the outer periphery of the chip, and the electric field strength near the bottom of the end trench 7 is relaxed.

  The distance W1 from the outer peripheral side wall of the end trench 7 to the inner peripheral end of the chip (position P in FIG. 1) in the region where the field plate 9 and the n-type drift layer 2 are in contact is from the position P to the guard trench. 8 is preferably smaller than the distance W2 to the end portion on the inner peripheral side of the chip. In this case, since the interval from the outer peripheral side wall of the end trench 7 to the position P is narrowed, the depletion layer that spreads at the time of reverse bias is pulled more strongly by the potential of the field plate 9. For this reason, since the depletion layer can spread toward the guard trench 8 with a low reverse bias voltage, the relaxation effect on the electric field strength near the bottom of the end trench 7 is enhanced. As a result, it is possible to prevent the avalanche current from being concentrated on the breakdown voltage structure 22. The relationship between W1 and W2 will be described later.

Next, a manufacturing method according to Example 1 of the present invention will be described. In the following, a TMBS diode with a rated voltage of 100 V is assumed.
The concentration of arsenic contained is 1 × 10 ¹⁹ / cm ³ or more, the thickness is 500 μm, and the mirror polished surface of the n-type semiconductor substrate 1 formed by the CZ method is the upper surface. On the upper surface of the n-type semiconductor substrate 1, an n-type drift layer 2 having a concentration of phosphorus contained of 4 × 10 ¹⁵ / cm ³ is deposited by an epitaxial growth method. Subsequently, 4000 mm of a thermal oxide film is grown on the upper surface of the n-type drift layer 2. Subsequently, patterning and etching (mainly anisotropic dry etching) are performed on the thermal oxide film by photolithography to form an oxide film mask for trench etching. Subsequently, silicon is etched from the opening of the oxide film mask by anisotropic etching to form a trench. Subsequently, a 3000 nm thermal oxide film is formed on the trench sidewall. Next, polysilicon doped with phosphorus is deposited by a chemical vapor deposition (CVD) method or the like. Subsequently, the polysilicon is etched so that the polysilicon 13 remains only inside the trench. Next, an interlayer insulating film such as BPSG (boron / phosphor glass) or HTO is deposited by a CVD method or the like. Subsequently, an interlayer insulating film is opened in a region where n-type drift layer 2 and anode electrode 3 or field plate 9 are connected by patterning and etching. Subsequently, a metal to be the anode electrode 3 is formed by a sputtering method or a vapor deposition method. The selection of the metal is performed using a barrier height determined by a Schottky junction between a known metal (such as molybdenum, titanium, tungsten, platinum, or palladium) and a semiconductor (such as silicon, silicon carbide (SiC), or gallium nitride (GaN)). In consideration of the rated voltage, this is done as appropriate. In Example 1, nickel was used. Subsequently, the anode electrode 3 is patterned and etched. Further, a polyimide film or a silicon nitride film is deposited, and patterning and etching are performed to form a passivation film (not shown). Next, back grinding is performed from the lower surface of the n-type semiconductor substrate 1 so that the remaining thickness including the n-type drift layer 2 and the n-type semiconductor substrate 1 becomes 300 μm. Subsequently, the cathode electrode 4 is formed on the grind surface by sputtering or vapor deposition. Finally, the wafer is diced by a diamond cutter or the like and cut into individual chips. In addition, about the order of the said process, you may replace a part in the range which can manufacture the Example of this invention.

  Here, the polysilicon 13 inside the trench is used to make the potential inside the trench the same as that of the anode electrode 3. Therefore, the material that fills the inside of the trench only needs to show conductivity, for example, aluminum and an alloy of aluminum and silicon, the same metal as the anode electrode 3, or a high melting point metal such as platinum having a melting point higher than that of silicon. I do not care. In the above description of the manufacturing method, the rated voltage is assumed to be 100V, but other rated voltages (30, 50, 200V, etc.) may be used. In this case, the thickness of the n-type drift layer 2, the doping concentration, the metal for the anode electrode 3, and the like may be adjusted or selected as appropriate.

  Next, the potential distribution and the cross-sectional electric field strength distribution when a reverse bias voltage of 100 V is applied between the TMBS diode of Example 1 of the present invention and the conventional TMBS diode are compared.

  FIGS. 5A and 5B show the distribution of equipotential lines 15 (also referred to as electrostatic potential distribution) in a cross section cut perpendicular to the chip surface when a reverse bias voltage of 100 V is applied. FIG. FIG. 5A is a conventional TMBS diode, and FIG. 5B is a TMBS diode according to the first embodiment. In particular, on the upper surface of the chip in FIG. 5-2, the finished dimensions on the chip surface of the active portion trench 12, the end trench 7, the guard trench 8, the interlayer insulating film 6, the anode electrode 3 and the field plate 9 (after completing the process) (This is a dimension in consideration of the etched portion from the dimension in the photomask). 5A is that the anode electrode 3 itself connected to the polysilicon 13 inside the end trench 7 has the function of the field plate 9 in the breakdown voltage structure 22 as well. That is, the field plate 9 of the breakdown voltage structure 22 is always at the same potential as the anode electrode 3. As a result, the equipotential lines 15 are densely distributed on the outer peripheral side of the bottom of the end trench 7, but the equipotential lines 15 are pulled in the direction of the outer periphery of the chip over the lateral length of the field plate 9. I understand. This alleviates the concentration of the equipotential lines 15 near the bottom of the end trench 7. On the other hand, an equipotential line 15 extending from 0 to 60 V enters the interlayer insulating film 6. This is because the relative dielectric constant (3.9) of the interlayer insulating film 6 made of silicon oxide is smaller than the value of silicon (11.9). Therefore, even if the length of the field plate 9 is sufficiently long as shown in FIG. 5A (about 16 μm in this drawing), the concentration of the equipotential lines 15 on the outer peripheral side of the bottom of the end trench 7 is still reduced. Not enough. On the other hand, in the structure of the first embodiment shown in FIG. 5B, the concentration of the equipotential lines 15 on the outer peripheral side of the bottom of the end trench 7 is reduced as compared with the conventional structure shown in FIG. ing. The potential in the vicinity of the bottom of the end trench 7 is also 45V, which is 15V smaller than that of the conventional structure. When the equipotential line 15 in the region (mesa region 18) of the n-type drift layer 2 between the end trench 7 and the guard trench 8 is viewed, it is curved toward the bottom (the lower surface direction) of the guard trench 8. I understand. This is because the field plate 9 connected to the inside of the guard trench 8 is also connected to the mesa region 18 so that the surface potential of the mesa region 18 is fixed to the potential inside the guard trench 8. In other words, the guard trench 8 pulls the potential of the mesa region 18 through the field plate 9, thereby reducing the concentration of the equipotential lines 15 at the bottom of the end trench 7.

The degree of relaxation of the equipotential lines 15 can be understood by looking at the electric field strength distribution. FIG. 6 is a diagram showing the electric field strength distribution along the cutting line when cutting from positions R1 to R2 in each of FIGS. 5-1 and 5-2. A thick line with a circular marker is the case of Example 1, and only a thin line is a conventional example. The electric field strength at the bottom of the trench at the position where the lateral distance is 5 μm, that is, the outer peripheral side of the chip of the end trench 7 is the maximum value. It can be seen that the maximum electric field strength of Example 1 is 4.3 × 10 ⁵ V / cm, which is about 14% lower than the value of the conventional field plate structure. The impact ionization rate due to avalanche is extremely sensitive to the electric field strength. For example, when the electric field strength increases by about 20%, the impact ionization rate becomes 5 to 10 times higher. Therefore, the reduction of the maximum electric field strength is extremely effective in preventing the avalanche from being generated by the breakdown voltage structure 22.

  Here, as shown in FIG. 5B, the width of the end trench 7 or the guard trench 8 is preferably larger than the width of the active trench 12. The active portion trench 12 itself does not serve as a current path when a forward current is applied, and thus becomes an ineffective region. Therefore, it is preferable to form the active portion trench 12 with the smallest width allowed by the process design rule. On the other hand, in the end trench 7, the end portion (active end portion 19) of the anode electrode 3 is terminated at the upper portion of the polysilicon 13 embedded in the end trench 7 as shown in FIG. Need to be. The guard trench 8 also needs to be terminated at the upper portion of the polysilicon 13 embedded in the guard trench 8. Therefore, the width of the end trench 7 or the guard trench 8 needs to be secured to some extent. If the width is larger than that of the active trench 12, the end of the anode electrode 3 or the field plate 9 can be stably terminated. Can do. Therefore, the width of the end trench 7 or the guard trench 8 is preferably larger than that of the active trench 12.

Next, how superior the structure of the first embodiment of the present invention is in terms of securing a withstand voltage as compared with the case where the conventional field plate structure is adopted as the withstand voltage structure of the TMBS diode will be described. FIG. 8A is a diagram showing a cross section from the active portion 21 to the withstand voltage structure portion 22 when the conventional field plate structure is adopted as the withstand voltage structure of the TMBS diode. As described above, in the TMBS diode, when the breakdown voltage structure portion 22 is formed without using the p-type guard ring layer 5, the active end portion 19 of the anode electrode 3 is replaced with the conductive polysilicon in the end trench 7. It is necessary to terminate at the inner region of 13. In order to reduce the electric field strength on the chip outer peripheral side at the bottom of the end trench 7, it is necessary to extend the length of the field plate 9 to the chip outer peripheral side. Therefore, from the center of the active portion trench 12 that is the closest to the termination trench 7 and a length of up to chip the outer peripheral end of the field plate 9 and L _E. A graph comparing the breakdown voltage of the conventional field plate structure when the _LE is changed with the breakdown voltage of the structure of Example 1 is shown in FIG. The longer L _E at conventional field plate structure, the breakdown voltage value L _E is 40μm or more as shown in Figure 8-2 is saturated 108V. This is because the equipotential lines is absorbed by the interlayer insulating film 6, since even extending the L _E above 40μm are not alleviated electric field strength. Therefore, L _E is 40μm is required at a minimum. On the other hand, in Example 1 formed with L _E of 20 μm, the breakdown voltage is 115 V, which can be 6% higher than L _E > 40 μm in the conventional field plate structure. Accordingly, even if the length of the pressure-resistant structure portion 22 is significantly reduced, it is possible to ensure a high breakdown voltage.

  Next, the point of the operation effect in Example 1 of the present invention will be described. The points of Example 1 of the present invention are as follows. That is, as described above, the surface of the mesa region 18 between the end trench 7 and the guard trench 8 is connected to the field plate 9 having the same potential as that of the conductive polysilicon 13 embedded in the guard trench 8. In other words, the potential of the mesa region 18 is fixed to the guard trench 8. The distance from the active end 19 to the chip inner peripheral end (hereinafter referred to as position P) of the region where the field plate 9 and the n-type drift layer 2 are in contact is W1, and the distance from the position P to the guard trench 8 The distance to the inner peripheral side end of the chip is W2 (all shown in FIG. 1). That is, W1 is the length of the region covered with the interlayer insulating film 6 on the surface of the mesa region 18. On the other hand, W2 is also the length of the region where n-type drift layer 2 is in contact with field plate 9 on the surface of mesa region 18. At this time, the relative relationship between W1 and W2 becomes important. FIG. 9 is a graph showing the relationship between the ratio of W1 and W2 (W2 / W1) and the element breakdown voltage. W1 is set to 2.0 μm. When W2 is smaller than W1, the withstand voltage is abruptly smaller than the value (115V) shown in FIG. The reason for this is as follows. When the length W2 of the region where the field plate 9 and the mesa region 18 are in contact with each other decreases, the effect of the guard trench 8 pulling the potential of the mesa region 18 through the field plate 9 is weakened. As a result, the equipotential lines 15 come out from the interlayer insulating film 6 to increase the electric field strength at the bottom of the end trench 7. Therefore, it is desirable that W2 is longer than W1. Further, when W2 is longer than W1, the withstand voltage increases to 119 V, and is substantially saturated when W2 is twice or more W1. That is, the length of W2 is twice as long as W1, and the effect of pulling the potential of the mesa region 18 is maximized and stabilized. Therefore, it is more preferable that W2 is at least twice W1.

  In an actual design, the mesa region 18 sandwiched between the end trench 7 and the guard trench 8 is mainly arranged in an annular shape on the surface of the chip. At this time, a part of the annular mesa region 18 may have a portion where W1> W2, and the actual withstand voltage is not required to be greatly reduced from the value of W1 <W2, as described above.

  When the mesa region 18 between the end trench 7 and the guard trench 8 is set to a floating potential without connecting the guard trench 8 to the n-type drift layer 2, there are the following problems. In the case where the mesa region 18 is floating in potential, when the depletion layer expands when a reverse bias is applied, the equipotential line 15 escapes from the oxide film 11 and the interlayer insulating film 6 formed on the side wall of the end trench 7 to the outside. The guard trench 8 cannot pull the equipotential line in the mesa portion. As a result, the state is substantially the same as the absence of the guard trench 8, and the electric field strength locally increases at the bottom of the end trench 7. As a result, an avalanche is generated at a low voltage value, and the breakdown voltage is lower than the breakdown voltage value of only the active portion 21. In addition, when the guard trench 8 that should be at a floating potential is charged for some reason when the trench is manufactured, the breakdown voltage is similarly reduced.

  FIG. 2 is a diagram showing a cross section of a TMBS diode according to Example 2 of the present invention. The difference from the first embodiment is that the active portion trench 12 is not formed in the active portion 21 and only the end trench 7 is formed. Since the active part trench 12 of the TMBS diode is not a current path in itself, it is an invalid region when conducting. For example, when the rated current is reduced (for example, 1 A or less) using a TMBS diode as a small current capacity application, the area of the active part 21 may be reduced, and the ratio of the area occupied by the active part trench 12 that is an invalid region may be increased. In that case, even if no trench is formed in the active portion 21, if the end trench 7 is provided as in the second embodiment, a sufficient breakdown voltage can be secured and a forward voltage drop can be reduced.

  Next, a TMBS diode according to Example 3 of the present invention will be described with reference to FIG. FIG. 3 is a cross-sectional view of the TMBS diode according to the third example. The difference of Example 3 from Example 1 is as follows. A p-type floating layer 10 is formed between the end trench 7 and the guard trench 8. The p-type floating layer 10 is in contact with the end trench 7. The p-type floating layer 10, the n-type drift layer 2 and the polysilicon 13 in the guard trench 8 are connected via the field plate 9. Since the depletion layer that spreads at the time of reverse bias can reach the p-type floating layer 10 before the guard trench 8, the effect of pulling the depletion layer (equipotential line) toward the outer periphery of the chip is further enhanced. As a result, the electric field strength in the vicinity of the bottoms of the end trench 7 and the guard trench 8 is further relaxed. A further important feature is that the p-type floating layer 10 is separated from the anode electrode 3 without being connected. That is, since the p-type floating layer 10 is not in contact with the anode electrode 3, the electric field strength can be relaxed without injecting holes, which are minority carriers, into the drift layer. Further, the depth from the upper surface of the n-type drift layer 2 in the p-type floating layer 10 may be shallower than the depth of the end trench 7 or the guard trench 8, but deeper than the end trench 7 as shown in FIG. It is preferable to do. When the junction depth of the p-type floating layer 10 becomes deeper than the end trench 7 or the guard trench 8, the depletion layer extends from the pn junction of the p-type floating layer 10. Therefore, the depletion layer hardly spreads at the bottom of the end trench 7 or the guard trench 8 at a position shallower than the junction depth. As a result, the electric field strength at the bottom of the end trench 7 hardly increases, and the breakdown voltage can be determined only by the structure of the active region.

FIG. 10A is a diagram illustrating a current-voltage curve when a reverse bias voltage is applied in the conventional field plate structure and the first and third embodiments of the present invention. Either here L _E is also a 20μm. When the voltage at which the avalanche current flows and the leakage current rapidly increases is defined as the withstand voltage, the withstand voltage of the conventional field plate structure is 104V, Example 1 is 115V, and Example 3 is 120V. A graph comparing the pressure resistance values is shown in FIG.

  Here, when the withstand voltage value is calculated with the structure of only the active portion 21, it becomes 115V. In other words, the withstand voltage of the conventional field plate structure is smaller than the ideal withstand voltage value determined only by the active portion 21, the same as the value of the active portion 21 in the first embodiment, and 5 V than the value of only the active portion 21 in the third embodiment. You can see that it is getting higher. This is an effect newly found in the third embodiment of the present invention, and this effect will be described below. Regarding Example 3, the electrostatic potential distribution when a reverse bias voltage of 100 V is applied is shown in FIG. 5-3, the potential of the mesa region 18 from the end trench 7 to the guard trench 8 is pushed to the outer periphery by the p-type floating layer 10, and the potential in the vicinity of the guard trench 8 is shown in FIG. It turns out that it is low compared with the structure of 1. That is, not only the potential of the mesa region 18 is pulled to the guard trench 8 side by the guard trench 8, but also the effect that the p-type floating layer 10 further assists the pulling of the potential of the mesa region 18 occurs. FIG. 7 shows the electric field intensity distribution on the cross section from the position R1 to R2 at this time. A thick line with a marker is Example 3. The electric field strength at the position where the lateral distance is 5 μm, that is, at the bottom of the end trench can be reduced to 60% as compared with the value at the same position of the conventional field plate structure. For this reason, it can be considered that the avalanche current no longer occurs in the breakdown voltage structure 22. Further, since the avalanche current generated in the active portion 21 is dispersed on the chip outer peripheral side which is the withstand voltage portion, the current density can be reduced as compared with the avalanche current of only the active portion 21. Since the avalanche breakdown is a positive feedback amplification phenomenon due to the avalanche current, in the case of the TMBS diode, the avalanche current is concentrated on a small cell pitch (about 3 μm) of the active portion 21, and amplification is likely to occur. On the other hand, when the breakdown voltage structure 22 where the avalanche current is no longer generated is adjacent to the avalanche current as in the third embodiment, the avalanche current is dispersed from the active portion 21 to the breakdown voltage structure portion 22, Alleviated. Therefore, the reverse bias voltage can be raised to a voltage slightly higher than the withstand voltage value of only the active portion 21. This is the reason why the breakdown voltage of Example 3 is higher than the breakdown voltage value of only the active portion 21.

  The provision of the floating p-type floating layer 10 as in the present invention has the following advantages. That is, the amount of increase in the breakdown voltage of the chip beyond the value of only the active portion 21 is reduced to increase the dopant concentration of the n-type drift layer 2 in the active portion 21 or to reduce the thickness of the n-type drift layer 2. It becomes possible to do. As a result, it becomes possible to obtain a TMBS diode having a lower on-resistance than can be obtained with a conventional withstand voltage structure.

  FIG. 4 is a cross-sectional view of a TMBS diode according to Example 4 of the present invention. The difference from the third embodiment is that the p-type floating layer 10 is formed so as to contact not only the end trench 7 but also the guard trench 8. In this way, the depletion layer is more likely to spread on the outer periphery of the chip, and as a result, the breakdown voltage value including the breakdown voltage structure portion 22 can be made larger than the breakdown voltage value of only the active portion 21.

Next, a method for treating the end portion in the longitudinal direction of the active portion trench 12 and a preferred positional relationship with the end portion trench 7 or the guard trench 8 will be described.
FIG. 11A is a plan view illustrating the structures of the active portion 21 and the breakdown voltage structure portion 22 according to the fifth embodiment. When this structure is viewed from an oblique direction, the structure shown in FIG. 11-2 is obtained. The active part trench 12 in Example 5 has a stripe shape. The longitudinal end of the stripe-shaped active portion trench 12 is connected to the end trench 7. On the other hand, an active curved trench 20 having a length shorter than the length of the active portion trench 12 is provided between the active portion trench 12 and the end trench 7. Both ends of the active curved trench 20 are curved at a radius smaller than the radius of curvature of the portion curved at the corner of the end trench 7, and both ends are provided at the end of the active trench 12. Connected to the trench. In this way, an end portion having a small curvature radius in the longitudinal direction of the trench as shown in FIG. 16A can be eliminated from the upper surface of the chip. Therefore, it is possible to sufficiently reduce the increase in electric field strength and cracks.

  Further, the processing method of the end portion in the longitudinal direction of the active portion trench 12 and the preferable positional relationship with the end portion trench 7 or the guard trench 8 may be as in the sixth embodiment of the present invention. FIG. 12A is a plan view of the structure of the active part 21 and the breakdown voltage structure part 22 according to the sixth example. When this structure is viewed from an oblique direction, a structure as shown in FIG. 12-2 is obtained. In the present structure, the active portion trench 12 has a donut shape as in the end trench 7 or the guard trench 8. Then, the active portion trench 12 is arranged so that the geometric center of gravity of the trench having the whole donut shape comes to the mesa region 18 surrounded by the innermost active portion trench 12 formed near the center of the chip. In this way, all the active part trenches 12 are terminated, and are adjacent to each other at equal intervals. Therefore, the electric field strength increases only at the corner position where the trench is curved (hereinafter referred to as a corner portion), and if the curvature radius is increased, the increase in the electric field strength can be ignored. If the radius of curvature of the active part trench 12 is, for example, equal to or greater than the pitch of the active part trenches 12 (unit period of repeated arrangement of adjacent active part trenches 12, hereinafter referred to as a trench pitch), electric field concentration at the corner part can be sufficiently suppressed. Further, the donut shape of the active portion trench 12 may be a shape in which a corner portion of a quadrilateral is chamfered as illustrated in FIG. 12A, or may be a circle (a perfect circle or an ellipse). It is sufficient that the small portion is about the above-described trench pitch. However, when the planar shape of the trench is circular, the area of the ineffective region increases in the vicinity of the four corners of the chip (quadron), so that the corner of the quadrangle is rounded as described above. Is desirable. Further, if the radius of curvature of the corner portion is 1000 times or less of the trench pitch, the proportion of the invalid region can be reduced to 3% or less, and the presence of the invalid region can be ignored. Further, the application of a preferable radius of curvature of the corner portion (equal to the trench pitch, that is, equal to or more than 1000 times) is not limited to the sixth embodiment. For example, the present invention may be applied to the corner radius of curvature of the trench in the vicinity of the end trench in the end trench 7 or the guard trench 8 or the active trench 12 in the fifth embodiment. By doing so, it is possible to exert an effect of suppressing electric field concentration and crack generation at the trench end in every trench.

  Here, regarding the fifth and sixth embodiments, the sectional shapes of the end trench 7 and the guard trench 8 will be described. FIG. 13A shows cross-sectional shapes from positions A to A ′ shown in FIGS. 11-1, 11-2, 12-1, and 12-2. FIG. 13-2 shows a cross-sectional shape from positions B to B ′ shown in FIGS. 11-1 and 11-2. The direction of the cross section of FIG. 13A is a cross section of a portion perpendicular to the longitudinal direction of the end trench 7 and the guard trench 8 aligned in parallel with each other, and has the same shape as the figure shown in FIG. On the other hand, FIG. 13-2 is a cross section at a position where the active trench 12 intersects the end trench 7 perpendicularly. In this case, although the active part trench 12 has a shape that continues in the longitudinal direction along the line BB ′, the end shape of the end trench 7 on the chip outer peripheral side is the same shape as that shown in FIG. Therefore, the mesa region 18 between the end trench 7 and the guard trench 8 and the connection form of the mesa region 18 with the field plate 9 and the polysilicon 13 inside the guard trench 8 are also the same as in FIG. It becomes. Therefore, regardless of the connection form of the active part trench 12 and the end part trench 7, the effect of the present invention can be obtained in the same manner, and it is electrically weak to a specific part by processing the end part of the active part trench 12. It can prevent that a part arises.

  Next, a TMBS diode according to Example 7 of the present invention will be described with reference to FIG. The difference of the seventh embodiment from the first embodiment is that an n-type surface layer 17 having a higher concentration than the n-type drift layer 2 is formed on the surface of the mesa region 18 between the end trench 7 and the guard trench 8. is there. The concentration of the n-type surface layer 17 is, for example, about 2 to 10 times that of the n-type drift layer 2. If the n-type surface layer 17 is present, the surface layer of the mesa region 18 may be charged when the charge enters the chip surface from the outside of the chip, and the distribution of the equipotential lines 15 may change. At this time, for example, a hole channel is formed in the surface layer of the mesa region 18, which causes an increase in leakage current, the electric field strength distribution changes, and a locally concentrated portion of the electric field strength is locally present on the surface layer of the mesa region 18. May occur and the withstand voltage may decrease. When the n-type surface layer 17 shown in the seventh embodiment of the present invention is formed, since the n-type dopant concentration in the surface layer of the mesa region 18 is high at the time of intrusion of external charges, charging is difficult to occur. For this reason, the formation of the hole channel or the change in the electric field strength distribution does not easily occur when an external charge enters, and the breakdown voltage or leakage current is stabilized. The above effect is sufficient when the maximum concentration of the n-type surface layer 17 is larger than that of the n-type drift layer 2 and is approximately twice or more. However, since equipotential lines cross the high concentration n-type surface layer 17, the distribution of equipotential lines also changes depending on the concentration. When the concentration of the n-type surface layer 17 is higher than 10 times that of the n-type drift layer 2, the electric field concentrates on the n-type surface layer 17 and the breakdown voltage is lowered. Therefore, the n-type surface layer 17 is preferably 10 times or less than the n-type drift layer 2. Further, when the depth of the n-type surface layer 17 becomes deeper than either the end trench 7 or the guard trench 8, the equipotential line distribution similarly changes and local electric field concentration tends to occur. Therefore, the depth is preferably shallower than either or both of the end trench 7 and the guard trench 8.

  Further, although not shown, in the structure shown in FIG. 3 of the third embodiment, in the mesa region 18 between the end trench 7 and the guard trench 8, an n-type surface layer is formed in a region where the p-type floating layer 10 is not formed. 17 may be formed. In this case, the n-type surface layer 17 is in contact with only the field plate 9 and is formed so as to be separated from the interlayer insulating film 6. In this way, the equipotential lines extend almost directly from the p-type floating layer 10 toward the guard trench 8, so that they do not cross the n-type surface layer 17. Therefore, the maximum concentration of the n-type surface layer 17 can be made higher than in the above case, and the breakdown voltage can be made more stable against external charges.

1 n-type semiconductor substrate 2 n-type drift layer 3 anode electrode 4 cathode electrode 5 p-type guard ring layer 6 interlayer insulating film 7 end trench 8 guard trench 9 field plate 10 p-type floating layer 11 oxide film 12 active portion trench 13 poly Silicon 14 Crack 15 Equipotential line 16 Schottky junction 17 N-type surface layer 18 Mesa region 19 Active edge 20 Active curved trench 21 Active portion 22 Withstand voltage structure

Claims (10)

A cathode layer comprising a semiconductor substrate of a first conductivity type;
A drift layer made of a first conductivity type semiconductor substrate having a lower concentration than the cathode layer is provided on one main surface of the cathode layer;
At least one first trench and an end trench surrounding the first trench are provided on an upper surface of the drift layer;
A first conductor is buried in the first trench and the end trench through an insulating film,
An anode electrode is provided on the upper surface of the drift layer so as to be in contact with the conductor and to form a Schottky junction with the drift layer,
In the semiconductor device in which a cathode electrode is provided on the other main surface of the cathode layer,
The outer peripheral end of the anode electrode is in contact with the first conductor of the end trench,
A field plate is provided apart from the anode electrode,
A second trench is provided so as to surround the end trench apart from the end trench;
A second conductor is embedded in the second trench through an insulating film,
The field plate semiconductor device characterized in that it connects to the surface the same potential of the drift layer in the mesa regions between the second conductor and said end portion and second trenches. The semiconductor device according to claim 1,
The distance W1 from the outer peripheral side wall of the end trench to the position P of the inner peripheral side end of the region where the field plate and the drift layer are in contact with each other,
A semiconductor device, wherein the distance is smaller than a distance W2 from the position P to an inner peripheral end of the second trench.   3. The semiconductor device according to claim 1, wherein a width of the end trench is larger than a width of the first trench. 4.   4. The semiconductor device according to claim 1, wherein the semiconductor device is disposed between the first trench and the end trench, and a length of the straight line portion is longer than a length of the first trench. The second end is curved at a radius smaller than the radius of curvature of the end trench 7, and the both ends are connected to the first trench provided at the end of the first trench. A semiconductor device having a trench. The semiconductor device according to any one of claims 1 to 3,
The first trench has a donut shape on the upper surface of the drift layer,
A geometrical center of gravity of the first trench having the donut shape is located inside the first trench formed on the innermost periphery of the first trench. The semiconductor device according to claim 1, wherein:
Connected to the end trench and / or the second trench;
And a second conductivity type floating layer connected to the field plate and formed on the upper surface of the drift layer is disposed so as to be separated from the anode electrode,
The depth of the floating layer from the upper surface of the drift layer is deeper than the depth of either or both of the end trench and the second trench. The semiconductor device according to claim 6.
The floating layer is in contact with both the end trench and the second trench.   8. The semiconductor device according to claim 1, wherein a concentration higher than a concentration of the drift layer is formed on a surface of the drift layer sandwiched between the end trench and the second trench. And a first conductivity type surface layer shallower than either or both of the end trench and the second trench.   9. The semiconductor device according to claim 8, wherein the maximum concentration of the surface layer is not less than a value indicated by the drift layer and not more than 10 times a value indicated by the drift layer. 6. The field plate according to claim 1, wherein the field plate forms a Schottky junction with the surface of the drift layer in the mesa region between the second conductor and the end trench and the second trench. The semiconductor device as described in any one